1912001 - NVDA will not read the Alt text out loud from an Added image of a Saved PDF

Status: VERIFIED FIXED

(Firefox :: PDF Viewer, defect, P1)

Description

[Rares Doghi, Desktop QA]

Found in

• Beta 130.0b1

Affected versions

• 131.0a1 (2024-08-07)
• Beta 130.0b1
• Release 129
• ESR 128

Affected platforms

• Windows

Preconditions
Have any screen reader enabled

Steps to reproduce

1. Open a PDF file in Firefox.
2. Add a few images to the PDF file.
3. Add some ALT text to each image.
4. Save the PDF file from the Toolbar.
5. Open the previously saved PDF in Firefox.
6. Reach the added images using the Tab key.

Expected result

• NVDA should read the Alt text from each image: "Figure + Alt text"

Actual result

• Reaching an Added image with Alt text, the Screen reader will read out loud "Page 1 Region clickable" instead of the Figure + Alt text.
  If there are 2 or more images, jumping with Tab to the next image NVDA will just read "Blank".
  Reaching back to an image with Shift + Tab will read the "Page 1 Region clickable" again.

Please note that the HTML tree shows the alt text if we look for it.

Regression range
Not a regression.

Comment 1

[Rares Doghi, Desktop QA]

I think this issue is related to the PDF we used, on the Dummy PDF file NVDA does not read the Alt Text, but from the sample5 file NVDA will read that Alt Text from the image.

Comment 2

[Rares Doghi, Desktop QA]

Attachment: dummy1.pdf

NVDA does not read the Alt text from this image : A person with long hair holding a large stick.

Comment 3

[Rares Doghi, Desktop QA]

Attachment: sample5.pdf

NVDA will read the Alt text from this image: Test Readers

Comment 4

[Rares Doghi, Desktop QA]

Attachment: dummy.pdf

There might be an issue with this PDF file but It also happened with others.
Adding images to these pdf files and saving them from the toolbar will not read the Alt text from the image. It will only say Blank or Page 1 region clickable.

Comment 5

[Calixte Denizet (:calixte)]

The file sample5.pdf is a tagged pdf which means that there is structure tree reflecting how the document is structured. In this case when we add an image we are able to correctly add our own tag at the right place in the tree.
But for the other pdf, it isn't tagged, so in such a case, when saving, we just create an empty structure tree where we just add our image with its alt-text.
That said when it's displayed in Firefox, we can probably have a workaround.

Comment 6

[Marco Castelluccio [:marco]]

(In reply to Calixte Denizet (:calixte) from comment #5)

The file sample5.pdf is a tagged pdf which means that there is structure tree reflecting how the document is structured. In this case when we add an image we are able to correctly add our own tag at the right place in the tree.
But for the other pdf, it isn't tagged, so in such a case, when saving, we just create an empty structure tree where we just add our image with its alt-text.

Why isn't NVDA reading it aloud? Shouldn't it be reading it, even if not exactly at the right place?

That said when it's displayed in Firefox, we can probably have a workaround.

I'm not sure it's worth it. After users edit PDFs, they expect the result to work exactly the same across viewers (at least the most frequent ones).

I think we should for now just make sure the image description is read aloud in untagged PDFs too, even if we can't put it in the right place, and then in the future when we have a way to generate tags automatically we can reconsider this.

Comment 7

[BMO Github Automation]

Attachment: [mozilla/pdf.js] [Editor] Make the stamp annotations alt text readable by either VO or NVDA (bug 1912001) (#18658)

Comment 8

[BugBot [:suhaib / :marco/ :calixte]]

The patch landed in nightly and beta is affected.
:calixte, is this bug important enough to require an uplift?

• If yes, please nominate the patch for beta approval.
• If no, please set status-firefox131 to wontfix.

For more information, please visit BugBot documentation.

Comment 9

[Rares Doghi, Desktop QA]

Verified as fixed in our latest Nightly build 132.0a1 (2024-09-15)

Comment 10

[Calixte Denizet (:calixte)]

We want to be sure that the patch won't induce any regressions, so it's a wontfix for beta.

Comment 11

[Rares Doghi, Desktop QA]

Updating the main status flag.

Comment 12

[Calixte Denizet (:calixte)]

This bug isn't specific to the new alt text flow so 130 is affected too.